#! /usr/bin/env python3
#
def montana ( ):

#*****************************************************************************80
#
## montana uses keras to solve a simple 1d regression case.
#
#  Discussion:
#
#    This is intended to be a short, simple, quick keras example.
#
#    The data is available as internal functions, and so this code
#    can be run standalone.
#
#    We are searching for the weights ( a, b ) in a formula:
#      y = a * x + b
#    that will best match a set of data.
#
#    We have 100 pairs of (x,y) data for training, and 10 pairs for testing.
#
#  Licensing:
#
#    This code is distributed under the GNU LGPL license.
#
#  Modified:
#
#    07 November 2019
#
#  Author:
#
#    John Burkardt.
#
#  Reference:
#
#    Francois Chollet,
#    Deep Learning with Python,
#    Manning, 2018,
#    ISBN: 9781617294433.
#
  import keras
  import matplotlib.pyplot as plt
  import numpy as np
  import platform
  
  print ( '' )
  print ( 'montana:' )
  print ( '  python version: %s' % ( platform.python_version ( ) ) )
  print ( '  keras version: %s' % ( keras.__version__ ) )
  print ( '  Neural network to solve a 1d regression problem.' )
  print ( '  Find weights (a,b) so that the formula' )
  print ( '    y = a x + b' )
  print ( '  best approximates a set of (x,y) data.' )
  print ( '  The data is read from internal functions.' )
  print ( '  There are 100 pairs of (x,y) data for training.' )
  print ( '  The resulting (a,b) are applied to 10 pairs of (x,y) test data.' )
#
#  Load the datasets.
#
  train = montana_train_data ( )
  print ( '' )
  print ( '  Training data shape: ', train.shape )
  train_data = train[:,0]
  train_targets = train[:,1]
  train_num = len ( train_data )

  test = montana_test_data ( )
  print ( '  Test data shape: ', test.shape )
  test_data = test[:,0]
  test_targets = test[:,1]
  test_num = len ( test_data )
#
#  Show the first three pairs of data and targets.
#
  print ( 'First 3 training items:', train_data[0:3] )
  print ( 'First 3 training targets:', train_targets[0:3] )
#
#  Create the keras model.
#
  from keras import models
  from keras import layers

  model = models.Sequential ( )
  model.add ( layers.Dense ( input_dim = 1, units = 1 ) )
  model.compile ( optimizer = 'sgd', loss = 'mse' )
#
#  Train the model.
#
  print ( '' )
  print ( 'Training the model:' )
  print ( '' )

  for step in range ( 101 ):
    cost = model.train_on_batch ( train_data, train_targets )
    if ( step % 10 == 0 ):
      print ( 'step %d: training cost = %g' % ( step, cost ) )
#
#  Test the model.
#
  print ( '' )
  print ( 'Testing the model:' )
  print ( '' )

  cost = model.evaluate ( test_data, test_targets, batch_size = test_num )

  W, b = model.layers[0].get_weights ( )

  print ( '' )
  print ( 'Results:' )
  print ( '' )
  print ( 'Test cost:', cost )
  print ( 'Weights = ', W ) 
  print ( 'Bias = ', b )
#
#  Plot the training data (blue) and testing data (red),
#  and the regression line (red).
#
  xmin = -1.0
  xmax = 2.0
  plt.plot ( [xmin, xmax ], [ b + W[0] * xmin, b + W[0] * xmax], 'r-' )
  plt.plot ( train_data, train_targets, 'bo' )
  plt.plot ( test_data, test_targets, 'ro' )
  plt.grid ( True )
  plt.xlabel ( '<--- X --->', fontsize = 16 )
  plt.ylabel ( '<--- Y --->', fontsize = 16 )
  plt.title ( 'Data: train (blue), test (red)', fontsize = 16 )
  filename = 'montana.png'
  plt.savefig ( filename )
  print ( '' )
  print ( '  Graphics saved as "%s"' % ( filename ) )
#
#  Terminate.
#
  print ( '' )
  print ( 'montana:' )
  print ( '  Normal end of execution.' )
  return

def montana_train_data ( ):

#*****************************************************************************80
#
## montana_train_data supplies the training data.
#
  import numpy as np

  data = np.array ( \
[[ 0.598499,    2.03206   ],
 [ 0.60241,     2.44061   ],
 [ 0.528659,    2.37146   ],
 [ 1.14069,     3.13872   ],
 [-0.229003,    1.25336   ],
 [ 1.25808,     3.95505   ],
 [ 1.65164,     5.08722   ],
 [-0.535303,    1.2575    ],
 [ 1.01164,     3.58375   ],
 [ 0.930335,    2.84445   ],
 [ 1.17088,     3.19438   ],
 [ 1.24618,     4.05287   ],
 [ 1.45564,     3.91006   ],
 [-0.706274,    0.636657  ],
 [-0.989338,    1.12846   ],
 [-0.307571,    1.35216   ],
 [-0.963235,    0.943437  ],
 [ 0.0606251,   1.25155   ],
 [-0.444156,    1.19095   ],
 [ 1.37831,     4.11453   ],
 [ 0.836516,    2.54402   ],
 [ 1.69829,     4.18295   ],
 [-0.0538967,   0.865774  ],
 [ 0.219966,    1.8477    ],
 [ 1.94078,     5.13762   ],
 [ 0.626196,    2.6553    ],
 [-0.273116,    1.41491   ],
 [-0.243275,    1.56301   ],
 [ 0.695467,    2.35417   ],
 [-0.599989,    0.996534  ],
 [ 1.62878,     5.12322   ],
 [ 0.440996,    1.95425   ],
 [-0.150022,    0.745887  ],
 [ 1.67663,     4.61879   ],
 [ 1.4829,      4.22713   ],
 [ 1.93991,     5.01128   ],
 [ 1.6017,      4.65738   ],
 [-0.419483,    1.5348    ],
 [ 0.363472,    1.45974   ],
 [ 1.41555,     4.0826    ],
 [ 1.24169,     3.51175   ],
 [ 1.75988,     4.41841   ],
 [-0.33859,     1.23152   ],
 [ 1.99386,     5.1235    ],
 [ 0.80839,     2.6468    ],
 [-0.893445,    1.26048   ],
 [ 0.666563,    1.96845   ],
 [-0.0410724,   1.74292   ],
 [ 0.506314,    2.31123   ],
 [ 0.599365,    2.24982   ],
 [ 0.98754,     2.53438   ],
 [ 0.374827,    1.70017   ],
 [-0.670371,    0.986843  ],
 [ 1.18981,     3.36155   ],
 [ 1.35261,     4.17052   ],
 [-0.576131,    1.34247   ],
 [ 1.51606,     4.36953   ],
 [ 0.208179,    1.84875   ],
 [ 0.219271,    1.76952   ],
 [ 0.111999,    1.60956   ],
 [ 1.75105,     4.69809   ],
 [ 0.472038,    2.18206   ],
 [ 1.89797,     5.0723    ],
 [-0.790837,    1.00822   ],
 [ 0.086146,    1.07676   ],
 [-0.86452,     0.845198  ],
 [ 0.00721245,  1.1904    ],
 [ 0.221532,    1.53029   ],
 [-0.300679,    0.789804  ],
 [-0.388771,    1.04663   ],
 [ 1.80441,     4.73272   ],
 [ 0.73114,     2.87377   ],
 [ 0.872665,    3.19095   ],
 [ 1.52891,     3.92502   ],
 [-0.470547,    0.812532  ],
 [ 1.55232,     3.90986   ],
 [-0.102594,    1.19779   ],
 [ 1.50326,     4.06359   ],
 [ 1.87327,     4.85862   ],
 [-0.655811,    1.04308   ],
 [-0.650234,    0.982951  ],
 [-0.862747,    1.05028   ],
 [ 0.744127,    2.63398   ],
 [-0.705918,    0.790914  ],
 [-0.360235,    0.819268  ],
 [-0.48344,     1.03121   ],
 [ 1.91945,     5.01807   ],
 [-0.405291,    0.995278  ],
 [-0.466338,    1.42394   ],
 [-0.184033,    1.73316   ],
 [-0.182296,    1.27067   ],
 [ 1.4159,      3.82413   ],
 [-0.48002,     0.977773  ],
 [-0.0765423,   1.68685   ],
 [ 0.396682,    2.00995   ],
 [ 1.4209,      3.93803   ],
 [ 1.45746,     3.89205   ],
 [-0.40407,     1.3084    ],
 [-0.542388,    1.27157   ],
 [ 1.06106,     2.95062   ]]
)

  return data

def montana_test_data ( ):

#*****************************************************************************80
#
## montana_test_data supplies the test data.
#
  import numpy as np

  data = np.array ( \
[[-0.748273,  0.950892],
 [-0.315139,  1.30002 ],
 [-0.312186,  1.13169 ],
 [ 0.492757,  1.55160 ],
 [ 0.833893,  2.80304 ],
 [-0.349539,  1.36128 ],
 [-0.802062,  1.18364 ],
 [ 1.046400,  3.37080 ],
 [ 0.758936,  2.15484 ],
 [-0.283329,  1.29814 ]]
  )

  return data

if ( __name__ == '__main__' ):
  montana ( )